Table of Contents
Imagine the following scenario: you’re writing code for your amazing new take-your-bike-to-work platform and you’ve just finished implementing a new feature to allow users to send unicorns to each other.
Your CI/CD pipeline has nicely tested, packaged and deployed the updates to your development Kubernetes cluster, you load the URL and are greeted by a very nice error page stating “Oops, my bad, we lost some unicorns”.
Clearly, you broke something, somewhere.
Most of the time, this means you’ll open up a terminal, run some commands to login into the cluster and start firing two dozen
kubectl-commands to figure out which microservice broke and check the logs to figure out where your code has broken.
You make some changes to the broken service and push your code to the repository and the CI/CD flow takes over again.
This process works quite nicely, but figuring which service is broken and which logs to check can be quite challenging.
kubectl-commands into the terminal probably takes half of the time you spend on debugging the issue.
As developers are always optimizing their workflow, using
kubectl just takes to much time, even with the
k alias for the command and perfect auto-complete features.
As the famous mantra goes: “anything worth doing twice is worth automating”. Therefore, quite some tools were created to make the process of navigating through a cluster easier than typing a lot of commands.
This blog post aims to provide a very brief overview of some of the more common tools that are available as replacements or additions to
kubectl to allow developers to look into a Kubernetes cluster.
All tools can be installed locally and don’t require any components to be installed in the cluster to operate.
K9s is a Kubernetes client built by Fernand Galiana.
The client is fully terminal-based so you’ll only be using your keyboard when operating it.
For those who are familiar with Vim, you’ll feel right at home in K9s.
It uses similar hotkeys to the popular editor.
There is a (quite steep) learning curve when you start using this client.
Once you have read the brief readme on the project’s home page and memorized the commands you’ll use the most, it is an absolute joy to use.
It feels like using
kubectl without the requirement to type all commands every time you need to get a deployment.
The tool is quite feature-rich at the time of writing.
You can port-forward, view secrets in plain text, edit resources directly, and “drill-down” from deployments into the logs of a container.
:deploy takes you to the pod overview,
<enter>-ing into the deployment takes you to all pods in that deployment.
<enter>-ing again into a pod reveals all containers, including (completed) init-containers.
<enter>-ing a final time into a pod takes you straight into a view with live logs.
This hierarchical approach feels very natural and follows the architectural design of Kubernetes.
A similar approach can be used for service (
:svc), statefulsets (
:sts) and deamonsets(
Another very familiar shortcut is the usage of
/ to filter on the context you’re currently in.
This works on basically any screen where you’d expect it, even in the logs view!
After a few days of use, it feels very natural to use. However, for those of us who rather use their mouse to navigate through resources and hate memorizing commands, this tool is not for you.
The project is still under very active development and quite some people are contributing to the codebase. Since the project doesn’t seem to be backed by a company directly, there are no real support guarantees nor is there a fixed release schedule. The maintainer however accepts fixes quite fast and releases are very frequent, sometimes multiple a day.
Demo: Go from deployment to all the way into pod logs
Demo: Switch between two Kubernetes contexts
Demo: Find out where a configmap is used
Octant is an open-source Kubernetes web dashboard built by VMWare Tanzu. It is written in Go, which is a trending programming language made by Google. It has lots of features, a clean user interface, and has the support of a big company in the IT industry behind it. This dashboard runs locally, which means you do not need to install it on your cluster, unlike the official Kubernetes dashboard.
Installing and accessing the dashboard is very straight-forward; it is as easy as running an install command (on Mac, we used Homebrew) and then running the
octant command from your terminal.
Once it’s running, you will find that the navigation is simple. We didn’t need to look around and search a lot to find what we need.
You are greeted by the application with a nice and well-ordered summary of your cluster resources. Navigating through the dashboard feels very comfortable and swift. They offer both a dark and light theme, which you will be able to see down below.
This client offers features that other traditional clients have as well such as log streams, a graphical display of all resources, switch between clusters,…
It does have unique features as well, such as in-app port forwarding (say goodbye to
Octant port forwarding
The most singular feature they have is their plugin system, which allows you to design plugins based on the information that you want on your interface. Although this requires knowledge of Go and their plugin system, you can customize your experience entirely to the way that you want it to be. For example, there is a Helm plugin available and Jenkins X has a repository of Octant plugins.
If you are a fan of web dashboards, then you should give Octant a try. Once you start to work with it, you will feel that it has a lot of potential. If you do prefer something more natively and working in a separate application window, then we suggest you to continue reading this blog post.
When first starting Lens, it immediately feels very easy to use.
Adding a cluster can be done by hitting the
+ and selecting a cluster from the dropdown.
Lens leverages the contents of the
kubeconfig it finds on the system to discover and authenticate with clusters.
This way, no additional magic is needed to get started.
After connecting to a cluster, you’re dropped into the cluster overview (see screenshot). This view provides you with an easy overview of the resources within the cluster and (super useful) provides a list of the last seen error events in the cluster.
Navigating to the list of pods provides an overview of all pods in the cluster.
By clicking on a pod you’re provided with the details of that pod (
From here, you can directly dive into the pod logs, shell into the pod, make edits or remove the pod from the cluster.
Similar support is available for most common resources within the cluster: statefulsets, deployments, configmaps, secrets, … The workflow is always the following: open the type in the sidebar on the left, click on an object to get details. From that detailed view, certain actions can be performed on the object.
As with most of the tools in this comparison, Lens is quite feature-rich. The most common types are supported and common actions are available. But Lens has another trick up its sleeve which makes it different from the other tools: Metrics/Prometheus integration.
The integration relies on a Prometheus instance being installed in the cluster that exposes the supported metrics. You can opt for Lens to install Prometheus (and other required components) for you, but in real scenarios, you either don’t have those rights or you’ll already have a Prometheus instance installed in the cluster. In those cases, you can just add some configuration to that instance and point the Lens app to that Prometheus instance. It will use port-forwarding under the hood, so no need to expose the Prometheus instance to the outside world. The integration works nicely and instantly provides some metrics about your cluster and deployed components. This provides good insights for developers to figure out their resource consumption without leaving their Kubernetes client. The charts and data seem to be very rudimentary, but improvements are expected to arrive over time.
As most of the clients described in this post, Lens app is an open-source project. Mirantis is behind the development of Lens, but at the time of writing, no supported (paid) version is available. There is continuous active development on the app and releases are about one month apart, so bug fixes and new features should be available regularly.
Screenshot: List of pods in Lens
Screenshot: Details about a pod in Lens, including Prometheus supplied metrics
Screenshot: Overview of a cluster in Lens, including the last error events
Kubenav is a rich featured, open source Kubernetes client created in early 2020 to manage your cluster(s) with. The application is under active development by the open-source community on their GitHub repository. There is cross-platform functionality, which means they provide you a desktop AND mobile client, which is a unique feature in the world of Kubernetes clients. They share the same codebase, so the navigation should be similar on all platforms.
The navigation menu is self-explanatory as it categorizes every Kubernetes resource, which you can then filter by namespace in the top-right corner. They support all widely used Kubernetes resources and the status and configuration of those resources. You can easily switch between Kubernetes clusters by using the drop-down menu available in the menu.
There is a window that is available at all times where you can consult different deployment logs and a terminal. Each window is categorized under a tab. When you minimize this window, you will notice a blue shell icon in the right bottom corner of Kubenav, which allows you to open your current logs or terminal again.
The app is made in the Ionic Framework using Capacitator as cross-platform runtime, which is something you notice when you start using the app. We personally don’t really like this feeling, as we prefer our clients to have a more native feeling, although we understand why the developers chose this approach.
The primary reason why we would use this application is because they support almost every resource combined with a GUI if you don’t like using the terminal. However, I’m not a fan of the interface in general, as it seems to be messy and overwhelming at times when a resource provides you with a great deal of information in an uncategorized way. If you want to know every detail about your resource at all times, then this application is definitely for you. We didn’t have a reason yet to use the mobile client, but we are sure it depends on your use case.
Infra App is a new addition to the list of Kubernetes clients. It is made by the people over at Docker Desktop & Kitematic and is being developed behind closed doors, which has been addressed as “unpleasant” within the Kubernetes community. It provides you with a clean, simplistic user interface that groups everything you need to know about a single resource together. Everything is self-explanatory and all the information you need is available within a few simple clicks.
When you open the application for the first time, you are greeted with a prompt asking you for your e-mail address. Although this is probably for newsletters and updates, we wish this step was optional.
You quickly notice that only basic functionality is available in the application, which makes sense as the client is still in early access at the time of writing. You can browse resources per namespace, go through application logs, read and edit YAML configurations, and check the current resources used by your deployment. There is a metrics interface for the whole cluster as well, which supplies you with a structured and detailed view about your nodes.
Since the application is still very young, it is lacking some functionality that you might expect or find in other clients. If you want something with more than basic functionality right now, this might not be the application you are looking for. However, if it has what you need, you will find that it will be very easy and straightforward to manage your Kubernetes cluster with this client.
Now the real question: which client should you use? As with any question about software, it depends. If you like to be lightning-fast and don’t mind struggling through a steep learning curve, K9s might be a tool for you. It’s the personal favorite of the authors this post, mainly because of its shortcuts and lightning-fast load times.
If you are using some software that has plugins available for Octant, definitely give it a try. One of the authors prefers Octant as non-terminal GUI. The plugins add a lot of value to the tool and might make it a very compelling option for your use-case.
If you often need to optimize your resource usage, want a client that just works, and is easy to use, go for Lens. This definitely hits the sweet spot between ease of use, stability, and available feature set.
If you need cross-platform functionality (especially mobile support), then Kubenav is the application you want. If you want to get used to the UI and dive into this software, we would suggest trying it on all platforms.
While Infra is still in early access, you can still use it in a production environment. If you like a simplistic yet structured interface with not too many features (yet), then Infra is the right choice for you.
But our final advice is: just try them out yourself and see which fits your workflow best. Most of them share the same basic functionality and it just depends on your use-cases and workflow which one fits best.